Interesting that this study accounted for regional differences between prostitution legalization regimes, definitely something that needs to be considered considering these kind of markets do not exist in a borderless vacuum.
Have to wonder if the scale effect dominating the substitution effect hints that perhaps the Swedish model is optimal for reducing trafficking wholesale while reducing harm (i.e. Johns need strong disincentives in order to not be terribly indifferent people in aggregate) They suggest that the Swedish model reduces the scale effect but not enough data to say if it increases the substitution effect.